Bold question:
I need to do some calculations on laminates of glass plies interleaved with plies of a viscoelastic interlayer. Plates are typically rectangular, edge length in the order of 2.0 m, supported at the 4 edges by a thick adhesive bead, and subjected to pressures of 60kPa to 100 kPa, near breakage.
Does anyone have any experience doing this with CalculiX?
Apologize for jumping in and bugging you all straight away. I did a forum search but my search terms did not return any results. When there are existing threads discussing this a pointer to these will probably get me started.
There are some special material models for brittle cracking that can be successfully used for laminated glass. One such model was developed by Hillerborg and is available in Abaqus. Another one was proposed by J. Pelfrene and is known as the crack delay model. The implementation of such advanced material models in CalculiX would require the use of UMAT subroutine.
If you don’t need to account for brittle cracking of the glass, standard models available in CalculiX (linear or nonlinear elasticity, Mises plasticity) can be sufficient.
Thanks for confirming my conclusion that CalculiX standard models would bring what is needed for large deformation cases of laminated glass. For now I want to evaluate the probability of first ply failure via the largest principal stress. I do know from people who do this now with general commercial apps like NASTRAN they ran into issues. Just wondering if anyone had gone the route before with CalculiX so I could use any experience shared.
One extension I think of in time is integration of the probability of failure of the surface elements. Reference
I would have to correct for the area of course. Pprobabilities shown are AFAIK for the area of a 400 mm diameter circle. I may need to write some code, bu that is phase 2, if ever it will pass.
Thanks for offering. I will ask for more details. What I understood until now the issues were mainly on finding the correct modeling of the interlayer and the adhesive. Under some conditions the shear deformation can amount to 2 or 3 times the thickness of the interlayer. The adhesive is subjected to combined compression, side translation and rotation and can see similar deformations.
To accurately model the interlayer and adhesive you would need a viscoelastic material model and cohesive elements, respectively. Those are available in Abaqus but unfortunately not in CalculiX. Thus, the material behavior will be represented less realistically. But you can utilize the hyperelastic material model and the results should still be quite good.
Thank you for responding. Good to hear you had (I understand) acceptable results.
Kuraray and Eastman provide extensive datasheets for their interlayers.
In the mean while I learned that getting reports of the relevant principal stress will be an issue. As far as I can see CGX default can report only the absolute value of the greatest principal stress, but for glass compressive stress is not relevant and I’d need to single that out. I went back to the theory implemented and will attempt to work from there.
Not really, the absolute maximum value is provided separately as worstPS but the three principal stresses in unmodified state are still available (as PS1, PS2 and PS3).
Until the first breakage strains are still small and a hyperelastic model for the interlayer is not necessary. PVB interlayers are typically very rate- and temperature sensitive though. But it may also not be necessary to include viscoelasticity in your material model, depending on how constant the conditions are.
For low speed / high temperature: interlayer is very flexible, the laminated glass panel acts more as 2 separate, sliding glass panes in bending
For higher speed / low temperature: interlayer acts stiffer, the laminated glass acts more as 1 solid (glass) panel with equivalent total thickness. Some authors actually model it this way to evaluate the principal stress (Rankine criterion) for breakage.
So if you have a more or less constant loading rate and temperature, you could just take a single Young’s modulus that matches those conditions (can be found in literature) and that’d be fine up until the first glass ply breaks.
To model the post-fractured response is a very different story. There are some different ways to model that, but all have their limitations.
Thank you for responding. I do fully agree on most points.
I really hope I can avoid using hyperelasticity. Note though that the glass material used is special grade chemically strengthened glass with a characteristic failure strength (CFS) in the order of 250 MPa as measured in a ISO1288-3 4-point bend test, so 0.072 m2 loaded area. This is about 6 times the CFS value of annealed glass. This may drive us into large deformation of the interlayer.
The interlayer properties we use are those for 25 deg. C and 10 seconds load duration, as declared by manufacturers.
